They are measuring the time until Vite is ready to serve requests, but then they compare it to the time of the webpack dev server, not mentioning that WebPack will have everything compiled at that time unlike Vite.

Vite is indeed much faster than webpack, but they use this highly misleading "vite ready to serve in 100 ms" stat.